NVIDIA GeForce GTX 1070 vs NVIDIA GeForce GTX 980 Comparison

Head-to-Head Benchmarks

The recorded data shows a decisive overall victory for the GTX 1070, which wins 10 of the 11 head-to-head comparisons. The most dramatic gap appears in 3DMark Steel Nomad DX12, where the GTX 1070 scores 1082 against 474 for the GTX 980, a 128.3% advantage. That is not a small margin; it is more than double the performance in a modern DirectX 12 workload, and it sets the tone for the rest of the analysis.

Compute-oriented tests also favor the newer card heavily. In Geekbench OpenCL, the GTX 1070 posts 44700 versus 34676, a 28.9% lead. PassMark GPU Compute shows a similar story: 6102 against 4753, a 28.4% gap. These results indicate that the GTX 1070 is not just faster in gaming scenarios; it also handles general-purpose compute workloads with noticeably more headroom. The GTX 980 is still a capable compute card, but the data suggests it trails by roughly a quarter to a third in these tasks.

DirectX 11 performance follows the same pattern. PassMark DirectX 11 scores 100 for the GTX 1070 and 83 for the GTX 980, a 20.5% delta. The older DirectX 9 test shows a 20.1% lead (197 vs 164), while DirectX 10 shows the largest legacy-API gap at 54.7% (82 vs 53). Even DirectX 12, where the GTX 980 is closest in the PassMark suite, still goes to the GTX 1070 by 4.3% (48 vs 46). That is a narrow margin, but it is a win nonetheless.

The Geekbench Metal benchmark, which targets Apple's Metal API, gives the GTX 1070 a 24% advantage (18801 vs 15163). PassMark G3D, a general 3D performance metric, shows a 21.7% lead (13498 vs 11095). Even the 2D test, PassMark G2D, goes to the GTX 1070 by 6.8% (846 vs 792), which suggests the newer card has a small but consistent edge even in less demanding rasterization tasks.

The single benchmark where the GTX 980 comes out ahead is Geekbench Vulkan. There, the GTX 980 scores 22543 against 22121 for the GTX 1070, a 1.9% margin. This is a narrow win, and it is the only test in the entire database where the older card takes the lead. It may reflect driver maturity or workload-specific behavior in that particular Vulkan test, but statistically it is within a rounding error of a tie. The overall picture remains clear: the GTX 1070 dominates the head-to-head results across almost every API and workload class.

Architecture Differences

The two cards come from different design generations, and the architectural gap explains most of the benchmark deltas. The GTX 1070 uses the GP104 chip built on Pascal architecture, manufactured on a 16 nm TSMC process. The GTX 980 uses the GM204 chip with Maxwell 2.0 architecture, built on a 28 nm TSMC process. That process shrink is significant: the GTX 1070 packs 7,200 million transistors into a 314 mm² die, giving a transistor density of 22.9M per mm². The GTX 980 has 5,200 million transistors on a larger 398 mm² die, with a density of only 13.1M per mm². The newer card fits more transistors into a smaller physical area, which is a direct contributor to its higher performance and lower power draw.

Clock speeds also reflect the architectural improvements. The GTX 1070 runs a base clock of 1506 MHz and a boost clock of 1683 MHz. The GTX 980 runs at 1127 MHz base and 1216 MHz boost. The GTX 1070 has a higher memory clock as well, with 2002 MHz (8 Gbps effective) versus 1753 MHz (7 Gbps effective) for the GTX 980. These clock differences are substantial, and they compound with the architectural efficiency gains of Pascal over Maxwell.

Memory configuration is another major split. The GTX 1070 comes with 8 GB of GDDR5 on a 256-bit bus, delivering 256.3 GB/s of bandwidth. The GTX 980 has 4 GB of GDDR5 on the same 256-bit bus, but only reaches 224.4 GB/s. The extra capacity matters for modern games and workloads that exceed 4 GB, while the higher bandwidth helps with texture streaming and high-resolution rendering.

The compute units tell a nuanced story. The GTX 980 actually has more shading units (2048 vs 1920) and more texture mapping units (128 vs 120), yet it still loses in almost every benchmark. Both cards have 64 ROPs. The reason for the GTX 1070's superiority lies in the architecture itself: Pascal's per-clock efficiency, higher clocks, and better memory subsystem overcome the raw unit-count deficit. The pixel rate for the GTX 1070 is 107.7 GPixel/s versus 77.82 GPixel/s for the GTX 980. Texture rate is 202.0 GTexel/s versus 155.6 GTexel/s. FP32 compute is 6.463 TFLOPS versus 4.981 TFLOPS. Those are the numbers that matter, and they all favor the GTX 1070.

Power consumption is notably lower on the newer card despite the performance advantage. The GTX 1070 has a TDP of 150 W, while the GTX 980 draws 165 W. The GTX 1070 also simplifies power delivery with a single 8-pin connector, whereas the GTX 980 requires two 6-pin connectors. Both cards suggest a 450 W power supply, but the GTX 1070 achieves more with less power, which is a direct benefit of the 16 nm process and Pascal's efficiency improvements.

Display outputs differ as well. The GTX 1070 offers 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a. The GTX 980 provides 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2. The newer DisplayPort revision on the GTX 1070 supports higher resolutions and refresh rates over that interface. Both cards support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4, so API-level feature support is effectively identical.

The physical dimensions are nearly identical: both are 267 mm long, 40 mm wide, and either 112 mm (GTX 1070) or 111 mm (GTX 980) tall. Both are dual-slot cards. If you are upgrading from a GTX 980, the GTX 1070 should fit in the same space without clearance issues.

The Verdict

The data points to the GTX 1070 as the clear choice in nearly every scenario. It wins 10 of 11 head-to-head benchmarks, often by large margins. The 128.3% lead in 3DMark Steel Nomad DX12 is the standout result, but the consistent 20% to 30% advantages across Geekbench OpenCL, PassMark G3D, and PassMark GPU Compute show that this is not a fluke. The GTX 1070 also uses less power (150 W vs 165 W) and has double the memory capacity (8 GB vs 4 GB), which makes it better suited for modern workloads and future titles.

The GTX 980's only win is in Geekbench Vulkan, where it leads by 1.9%. That is a narrow margin and unlikely to matter in real-world use. If you are building a new system or upgrading from an older card, the GTX 1070 is the better investment strictly from the performance data. The GTX 980 still holds up in legacy DirectX 9 and 11 workloads, but it trails by 20% or more in those same tests. The GTX 1070 offers higher clocks, more memory, better bandwidth, and lower power consumption. There is no benchmark category in the database where the GTX 980 is the recommended pick except for the single Vulkan test, and even there the difference is within noise.

FAQ

Q: Which card is faster in DirectX 12?

A: The GTX 1070. In 3DMark Steel Nomad DX12, it scores 1082 versus 474 for the GTX 980, a 128.3% lead. In PassMark DirectX 12, the GTX 1070 wins by a smaller 4.3% margin (48 vs 46).

Q: Does the GTX 980 win any benchmark?

A: Yes, one. In Geekbench Vulkan, the GTX 980 scores 22543 against 22121 for the GTX 1070, a 1.9% advantage. It loses the other 10 head-to-head tests.

Q: How much more memory does the GTX 1070 have?

A: The GTX 1070 has 8 GB of GDDR5, while the GTX 980 has 4 GB. Both use a 256-bit bus, but the GTX 1070 has higher bandwidth at 256.3 GB/s versus 224.4 GB/s.

Q: Which card consumes less power?

A: The GTX 1070, with a TDP of 150 W compared to 165 W for the GTX 980. The GTX 1070 also uses a single 8-pin power connector, while the GTX 980 requires two 6-pin connectors. Both recommend a 450 W power supply.

Q: Are the cards the same physical size?

A: Nearly identical. Both are 267 mm long and 40 mm wide. The GTX 1070 is 112 mm tall, and the GTX 980 is 111 mm tall. Both are dual-slot cards.

Q: Do they support the same APIs?

A: Yes. Both support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The display outputs differ: the GTX 1070 has DisplayPort 1.4a, while the GTX 980 has DisplayPort 1.2.

Where Each One Wins

The GTX 1070 wins in almost every category, but the size of the win varies. Its biggest advantage is in modern DirectX 12 workloads, where the 3DMark Steel Nomad test shows a 128.3% lead. That makes it the clear pick for anyone running current AAA titles or future releases that leverage DirectX 12. It also wins decisively in compute-heavy tasks like Geekbench OpenCL (28.9% lead) and PassMark GPU Compute (28.4% lead), which matters for users who do rendering, machine learning inference, or other GPU-accelerated work. The GTX 1070 is also the better choice for high-resolution gaming because of its 8 GB memory buffer and 256.3 GB/s bandwidth, both of which exceed the GTX 980's 4 GB and 224.4 GB/s.

The GTX 980's only clear win is Geekbench Vulkan, where it leads by 1.9%. That is a narrow margin, and it does not suggest any broad Vulkan superiority. If you are running a specific Vulkan-based workload that is highly sensitive to that benchmark's characteristics, the GTX 980 might match or slightly exceed the GTX 1070, but the difference is small enough to be considered negligible in practice. The GTX 980 also retains some relevance in legacy DirectX 9 and DirectX 10 applications, but it still loses those tests by 20.1% and 54.7% respectively, so "relevance" is relative.

For users upgrading from a GTX 980, the data strongly suggests the GTX 1070 is a worthwhile jump. The 21.7% lead in PassMark G3D and the 24% lead in Geekbench Metal translate to real-world gains in most games and applications. For users on older cards, the choice is equally clear: the GTX 1070 is the better performer across nearly the entire benchmark suite, with lower power draw and double the memory. The GTX 980 is not a bad card, but the recorded data leaves little room for debate about which one is the superior product.

Specification Differences

The two cards differ in several key specification fields. The GTX 1070 uses the GP104 chip with Pascal architecture, while the GTX 980 uses GM204 with Maxwell 2.0. The process node is 16 nm for the GTX 1070 and 28 nm for the GTX 980. Transistor counts are 7,200 million versus 5,200 million. Die size is 314 mm² versus 398 mm². Transistor density is 22.9M per mm² versus 13.1M per mm². Base clocks are 1506 MHz versus 1127 MHz, and boost clocks are 1683 MHz versus 1216 MHz. Memory clocks are 2002 MHz (8 Gbps effective) versus 1753 MHz (7 Gbps effective). Memory size is 8 GB versus 4 GB. Memory bandwidth is 256.3 GB/s versus 224.4 GB/s. Shading units are 1920 versus 2048. TMUs are 120 versus 128. ROPs are 64 on both. Pixel rate is 107.7 GPixel/s versus 77.82 GPixel/s. Texture rate is 202.0 GTexel/s versus 155.6 GTexel/s. FP32 compute is 6.463 TFLOPS versus 4.981 TFLOPS. FP16 is listed for the GTX 1070 at 101.0 GFLOPS (1:64), while the GTX 980 has no FP16 data. TDP is 150 W versus 165 W. Power connectors are 1x 8-pin versus 2x 6-pin. Display outputs are 1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.4a versus 1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.2. Height is 112 mm versus 111 mm. The GTX 1070 has a launch MSRP of 379 USD, and the GTX 980 has a launch MSRP of 549 USD. Both are end-of-life products, and both support the same API set.